<![CDATA[Abstract. The rapid advancement of artificial intelligence has enabled the creation of deepfake content that is increasingly realistic and difficult to distinguish from authentic media. The widespread dissemination of deepfakes on social media, particularly TikTok, has generated diverse public responses due to risks of misinformation, opinion manipulation, and ethical concerns. As one of the most popular platforms in Indonesia, TikTok provides a space for public opinion through user comments. This study aims to analyze public sentiment toward deepfake content on TikTok using a deep learning approach based on the IndoBERT model. The dataset consists of 4,000 Indonesian-language comments collected through web scraping using deepfake-related keywords. Sentiment labeling was conducted using a lexicon-based approach and classified into positive, neutral, and negative categories. The research process includes text preprocessing, data splitting, and IndoBERT fine-tuning. Model performance is evaluated using accuracy, precision, recall, and F1-score. The results show that negative sentiment dominates, indicating public concern over potential misuse of deepfake technology. This study provides insights for policymakers and platform stakeholders.]]>
Copyrights © 2026
